EEVblog Electronics Community Forum
Electronics => RF, Microwave, Ham Radio => Topic started by: jujun on May 31, 2018, 01:50:40 pm
-
Hello,
If I make two band pass, one for UHF and one for VHF, then I connect them together and I add a connector between them.
UHF Ant <==> UHF Band pass <==[SMA]==>VHF Band Pass <===> VHF antenna
Will the input be at 50 ohms if my two filter have inputs/outputs at 50 ohms ?
Will it be a diplexer ?
It will be OK to trasmit with it ?
Thank you
J
-
If like this, then yes
http://kw4fb.com/micro-diplexer/ (http://kw4fb.com/micro-diplexer/)
-
If like this, then yes
http://kw4fb.com/micro-diplexer/ (http://kw4fb.com/micro-diplexer/)
Ok, but this one have low pass, high pass filter, no band pass filter. So it dosent matter ?
-
No, it depends on the filter type. They need to be complementary, and wired in a compatible way.
If one filter has a high impedance, out of band (incident radiation reflects in phase), then another filter (with the same behavior) can be connected in parallel (with its input port).
If one filter has a low impedance, out of band (incident radiation reflects out of phase), then another filter (with the same behavior) can be connected in series (with its input port).
Series connection of ports is probably undesirable, because you need to worry about the common mode, ground isolation or something like that. May be perfectly alright for a coupled-resonator type filter, where isolation is normal. For a wideband, lumped element filter, not so much.
The constraints can be expressed more precisely, like adjacent bands being self-dual, i.e., Butterworth characteristic (otherwise, Cheby. for example has dips and peaks in its input impedance, which therefore need to be connected in series AND parallel at the same time, which is impossible -- i.e., it is not self-dual).
If you are leaving space between bands, then merely accounting for the asymptotic impedance of the other filter, in the design of each filter, will be okay.
On a practical note, notice this assumes zero transmission line length between filters and connections. Adding that, complicates things further. (Because a 1/4 wave TL can be used to convert a shorting-mode filter to an open-mode filter, and vice versa; but only over the bandwidth where the TL looks 1/4-wave-ey, so this is only useful for narrow band filters.)
If you have the option of building a new antenna completely, you may find it's better to use an "ultrawideband" design, like a bowtie, conical dipole, log periodic or such. Hacking together existing antennas, that, I guess obviously, aren't suited to your overall purpose, is probably not a good idea. Doable, but you need instrumentation to prove it works.
Tim
-
First, thank you for taking time to answser.
No, it depends on the filter type. They need to be complementary, and wired in a compatible way.
So, with the standard way of building a diplexer using a LPF and a HPF, it works because they are complementary (also in term of impedance VS frequencies)
If you are leaving space between bands, then merely accounting for the asymptotic impedance of the other filter, in the design of each filter, will be okay.
What I wanted to build is a BPF of 135Mhz to 150Mhz for VHF and a BPF of 430Mhz to 440Mhz for UHF.
For the VHF one I will make a box with helical resonators, and for the UHF one I will make a combline filter (with RG402 as variable capacitor) (I already build one for an another frequency)
The goal for these filters is to have very low losses, so I can put an LNA just after. I am also worried about the LNA oscillating if the matching is not good. Then I connect an RTL SDR. This two filters will enable me to monitor both UHF and VHF, without every annoying signal overloading my SDR, so I will be able to hear very faint signals.
If you have the option of building a new antenna completely, you may find it's better to use an "ultrawideband" design, like a bowtie, conical dipole, log periodic or such. Hacking together existing antennas, that, I guess obviously, aren't suited to your overall purpose, is probably not a good idea. Doable, but you need instrumentation to prove it works.
It's already antennas that I have built. It's yagi antenna. I am building a satellite ground station. (I am a new Ham ...)
About instrumentation, it's not a problem, because I can go to the ham radio club of my city, and there is a VNA up to 3Ghz, and a spectrum analyser up to 40Ghz.
Sorry, English is not my native language.
-
Ah, so your antennas are already pretty narrow band (and high gain is required), so you could actually do it with TLs and such. And, you sorta-kinda are with the helical resonators, and exactly are with the combline, so there we have it. :)
So yeah, it depends what the characteristics of the filters are. Ideally, you would measure the system by measuring the radiation from the two antennas, while driving the feedpoint. The filters are correct when the feedpoint SWR is small within each band, and only the intended antenna is radiating.
You can't generally ensure that radiation comes from the correct one, or that the filters will work with the antennas, or with each other, without evaluating the reactances from all these components, across the spectrum.
That is to say, you can't avoid the fact that, you've got complex numbers spinning around, and when you put two and two together, you get yet another complex number, and in general, their result won't be a well-behaved real number (resistance).
There are some strategies to help out, here -- if you use diplexing filters throughout, and terminate the unused ports, then the filter's output impedance becomes a constant resistance. The output port sees 50 ohms (from the terminator) when out-of-band, and 50 ohms (from the receiver) when in-band. Then, because the antenna sees 50 ohms at all frequencies, its response will be the same as when it was tested in a 50 ohm system (say with a VNA and another antenna to measure gain and pattern).
Likewise with the receiver side, the filters can be designed as multi-stage diplexers, so the high, middle and low stop-bands get terminated, while the high and low pass-bands get passed to their respective antennas.
This is probably an over-the-top, belt-and-suspenders method, but it has the least sensitivity to reactive sources and loads.
Now, with instrumentation, you should be able to build the individual filters normally, and determine their stop-band impedance (as measured at the plane where the filters would be joined -- because again, line length matters, including BNC tees and stuff). Once you know this, you can adjust that port (with L or C, resonators, TL segments, traps/stubs, whatever) to get a high impedance, and then it should be okay to connect the filters together. Then, repeat the measurements to verify that they're behaving as expected (and to tweak them further).
The same process (ensuring complementary impedances) applies to the filter-to-antenna connection as well. If nothing else, at least verify that the filter-antenna pair is working as intended.
Then repeat the measurements with the antennas attached to the joined filter pair, and you'll have the full system verified and working.
Your English is quite good :)
Tim
-
I'd just put it together with some short transmission lines between the common port and filters and then change the lengths of the lines until it works ok.